HomeNetRescue: um serviço SDN para detecção e solução de problemas em redes domésticas

Abstract

Home networks have become increasingly large due to the number of connected smart devices. Indeed, the use of personal devices, like smartphones, tablets, smart TVs, results in a densification of such networks. In addition, new services such as streaming, cloud storage, and torrents have become popular. Therefore, the need for greater reliability and Quality of Service (QoS) arises, which requires a more efficient management of home networks. In these networks, usually when a problem occurs, the users are not able to inspect or troubleshoot it. Moreover, such users are usually responsible for setting up their own network. This situation is aggravated by the lack of tools that support the identification of the problem source, as well as its automatic solution. In this work, we propose HomeNetRescue, a software defined network service for home networks autonomous management. HomeNetRescue is designed for the detection, diagnosis and automatic solution in wired and wireless networks. Its architecture is generic and modular, which allows the addition of new applications and devices to be monitored. The service was modeled to be used by Internet Service Providers (ISP) to manage problems in their customers' home networks. In addition, the service can add automatic detection and troubleshooting capabilities to these networks, thus the networks become more stable and reliable. The use of HomeNetRescue can provide cost savings for the ISPs, generate lower demand for support services, and shorten recovery time in case of failures. We evaluate the prototype of HomeNetRescue in a real environment. In order to do that, we elaborated three use cases. With our results, we can conclude that HomeNetRescue provided individual and global benefits on the network. These benefits correspond to gains in the throughput, and reductions in delay and jitter of the stations' wireless transmissions. In first evaluation we describe two approaches for transmit power control of access points, one distributed and one centralized. In Uncontrolled Transmit Power Control use case, the service detects an increase in the amount of packet loss due to collisions, acts automatically on the network, thus increases transmission power, and combats bad wireless links. This results in a gain of 7% in the throughput of the analyzed access point. In Transmit Power Control with Coordination use case, the service improved the average throughput by 66%, reduced the delay by 36% and the jitter by 15% when compared to a network without power control. Finally, in Dynamic and Coordinated Selection of Channels use case, the service performs an orchestrated assignment of channels to network access points in a multi-AP environment. In this case, the throughput improves by 131%, and delay and jitter reduces by 46% and 24% respectively, when compared to algorithms for channel assignment of commodity access points.